Vol. 78 CHROMATOGRAPHY OF CEREBROSIDE AND SULPHATIDE 185

2. Chromatography of the washed, deprotein-ized, lipid extract on alumina with a gradient-elution technique, in which the water content ofthe solvent was slowly increased, led to the suc-cessive elution ofcerebroside, sulphatide and amino-phospholipid. The cerebroside and sulphatide peaksalways overlapped. With one batch of alumina thegaI6ctose-containing lipids were completely separ-ated from the aminophospholipid.

3. Gradient-elution chromatography on silicicacid, whereby the methanol content of the solventwas increased, led to the complete separation ofthe cerebroside from the suiphatide. The formerwas also separated from the aminophospholipid,whereas there was overlapping between thesulphatide and aminophospholipid.

REFERENCES

Berenblum, I. & Chain, E. (1938). Biochem. J. 32,295.

Bruckner, J. (1955). Biochem. J. 60, 200.Davison, A. N., Dobbing, J., Morgan, R. S., Wajda, M. &

Payling-Wright, G. (1960). In Biochemistry of Lipids,p. 85. Ed. by Popjik, G. London: Pergamon Press Ltd.

Drake, B. (1955). Ark. Kemi Min. Geol. 8, no. 1.Folch, J., Arsove, S. & Meath, J. A. (1951). J. biol. Chem.

191, 819.Folch, J. & Lees, M. (1951). J. biol. Chem. 191, 807.Folch, J., Lees, M. & Sloane-Stanley, G. H. (1957). J. biol.Cerm. 226, 497.

Garton, G. A. & Duncan, W. H. R. (1957). Biochem. J. 67,340.

Gelmo, P. (1906). Ber. dtsch. chem. Ges. 39, 4175.Gray, G. M. (1958). Biochem. J. 70, 425.Gray, G. M. & Macfarlane, M. G. (1958). Biochem. J. 70,

409.

Hanahan, D. J., Dittmer, J. C. & Warashina, E. (1957).J. biol. Chem. 228, 685.

Hanahan, D. J. & Jayko, M. E. (1952). J. Amer. chem. Soc.74, 5070.

Jones, A. S. & Letham, D. S. (1956). Analyst, 81, 15.King, E. J. (1946). Micro-analysis in Medical Biochemistry,

1st ed., p. 17. London: J. and A. Churchill and Co. Ltd.King, E. J. & Wootton, I. D. P. (1956). Micro-analysis in

Medical Biochemistry, 3rd ed., p. 16. London: J. and A.Churchill and Co. Ltd.

Lea, C. H. & Rhodes, D. N. (1953). Biochem. J. 54, 467.Lea, C. H., Rhodes, D. N. & Stoll, R. D. (1955). Biochem. J.

60, 353.Lees, M. B. & Folch, J. (1959). Biochim. biophys. Acta, 81,

272.Lees, M., Folch, J., Sloane-Stanley, G. H. & Carr, S. (1959).

J. Neurochem. 4, 9.Long, C. (1943). Biochem. J. 37, 295.Long, C. & Penny, I. F. (1957). Biochem. J. 65, 382.Long, C., Shapiro, B. & Staples, D. A. (1960). Biochem. J.

75, 17P.Long, C. & Staples, D. A. (1959a). Biochem. J. 73, 7P.Long, C. & Staples, D. A. (1959b). Biochem. J. 73, 385.Long, C. & Staples, D. A. (1960). Biochem. J. 75, 16P.Lovern, J. H., Olley, J., Hartree, E. F. & Mann, T. (1957).

Biochem. J. 67, 630.Moore, S. & Stein, W. H. (1948). J. biol. Chem. 176, 367.Payne, S. & Platt, B. S. (1958). Proc. Nutr. Soc. 17, xvi.Radin, N. S. & Brown, J. R. (1960). Biochemical Prepara-

tion8, 7, 31. Ed. by Lardy, H. A. New York andLondon: John Wiley and Sons Inc.

Radin, N. S., Brown, J. R. & Lavin, F. B. (1956). J. biol.Chem. 219, 977.

Rhodes, D. N. & Lea, C. H. (1957). Biochem. J. 65, 526.Robins, E., Lowry, 0. H., Eydt, K. M. & MoCaman, R. E.

(1956). J. biol. Chem. 220, 661.Shapiro, B. (1953). Biochem. J. 53, 663.Svennerholm, L. (1956). J. Neurochem. 1, 42.Weiss, B. (1956). J. biol. Chem. 223, 523.

Biochem. J. (1961) 78, 185

The Phospholipids of Butter and theirEffect on Blood Coagulation

BY J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGANDepartment of Chemical Pathology, We8tminster Medical School, London, S.W. 1

(Received 4 May 1960)

The suggestion by Macfarlane (1955) that dietaryfats may play an important role in the aetiology ofatheroma and that different lipids may have widelydifferent effects on blood thromboplastin formationled us to investigate the effect of a number ofdietary fats on blood coagulation in vitro. At thestart of this work it was generally believed that thelecithin group of phospholipids were strong poten-tiators of blood coagulation (Macfarlane, Trevan &

Attwood, 1941; Fullerton, Davie & Anastasopoulos,1953) but later work by a number of workersshowed that phosphatidylethanolamine preent inthe crude lecithin was the active material, and thatpure lecithin was completely inactive (Poole,Robinson & Macfarlane, 1955; Lea, Rhodes &Stoll, 1955; Lea, 1956; O'Brien, 1956a, b; Robinson& Poole, 1956). Poole & Robinson (1956) havesince demonstrated that synthetic specimens of

J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGAN

phosphatidylethanolamine were as active, in rat-plasma systems, as that isolated from eggs. On theother hand, Biggs & Bidwell (1957) in their experi-ments on the fractionation of brain phospholipidsfound the blood-clotting activity variably distri-buted through the fractions and their phosphatidyl-ethanolamine fraction appeared to be the leastactive of those tested. They further found that syn-thetic dimyristoylphosphatidylethanolamine wasinactive in human plasma even in the presenceof egg lecithin as an emulsifier. The activity ofsynthetic phosphatidylethanolamine was slight inrat-plasma systems in comparison with human-brain phosphatides.

In the present work, the clotting activity of alarge number of fats and lipid fractions had to betested and hence a rapid blood-clotting techniqueinvolving a fat-free thromboplastin was required.The choice of test was therefore limited to theplasma Stypven (Russell viper venom, R..v.v.)method as used by Macfarlane et al. (1941), venombeing one of the few thromboplastins which is freeof fat. A modification of the above method, the'instantaneous' technique (Billimoria, Lockey &Maclagan, 1957), was preferred, since with thistechnique the lipolytic action of venom on plasmafats was minimized, and interfering effects of lipidsproduced by lipolysis were thus reduced.The effect of a number of the usual dietary fats on

this system showed that the clotting effect wasalmost entirely confined to milk fats and wasmaximal in butter (Maclagan & Billimoria, 1956).The activity was associated with the lipoprotein inbutter, and further separation indicated that theactive materials were phospholipids. This paperdescribes the preliminary separation of the phos-pholipids from butter and the further fractionationof these lipids by continuous gradient elution fromsilicic acid columns. The separation was controlledthroughout by tests of thromboplastic activitysince we were mainly interested in fractions withthis action. A preliminary account of some of thework has been published (Billimoria & Curtis,1959).

METHODS

Total nitrogen. This was determined by nesslerization bythe method of King & Wootton (1956) with a modifieddigestion mixture (C. Long, personal communication)consisting of 100 ml. of H2SO4, 25 g. of K2SO4, 25 mg. ofHgO and water to 21. Digestion of the lipid containingusually 10-20,ug. of N was carried out with 0-5 ml. of thedigestion mixture heated for 20 min. on an electric micro-digestion apparatus (A. Gallenkamp and Co. Ltd., London,E.C. 2).Amino nitrogen. This was determined on the intact

lipids by the procedure of Lea & Rhodes (1954); prelimin-ary hydrolysis with 0-5N-HCI at 1000 for 15 min. wasemployed for the glycosides.

Phosphorus. This was determined after removal of thesolvent followed by digestion with perchloric acid by aslight modification of the method of Allen (1940) in whichthe volume used in the colorimetric determination wasdecreased in order to increase the sensitivity of the method.Perchloric acid (0-5 mi., 60%, w/v) was used for the di-gestion of lipid containing 2-50 eg. of P. The digest wasdiluted with 5 ml. of water followed by the addition of0-25 ml. of 8% (w/v) ammonium molybdate and 0 5 ml. ofamidol reagent made up by dissolving 1 g. of amidol in100 ml. of water containing 18-3 g. of sodium metabisul-phite. The blue colour obtained was read after 12 min. in aUnicam SP. 600 spectrophotometer at 690 m,. in a 1 cm.cuvette.

Sugars. These were estimated by the orcinol-H2SO4method cited by Winzler (1955).Paper chromatography of bases and sugars formed on

hydrolysis. For bases, a sample of the combined peakfractions, obtained by chromatography, containing about5 mg. of the lipids was introduced into a thick-walledPyrex tube; the solvent was evaporated in vacuo and 1 ml.of 6N-HCI added. The tube was sealed and heated over-night at 1000. The tube was cooled and cut open and theHCI evaporated off in vacuo (0-1 mm. Hg). Water (0 5-1 ml.) was added and the solution evaporated again toremove traces of HCI. The residue was dissolved in 0-1-0-2 ml. of water, and 5pul. of the solution was applied to thepaper.For chromatography of sugars, samples containing 5 mg.

of sugar were hydrolysed for 2 hr. at 1000 with 2 ml. ofN-H2SO4. An excess of BaCO3 was added, and the mixturewas heated to boiling and filtered. The precipitate waswashed with 5 ml. of water, and the filtrate was evaporatedto dryness in vacuo (0-1 mm. Hg). The residue was dis-solved in 0-5 ml. of water, and 5-10l l. was applied topaper for chromatography.For chromatography of phosphate esters, samples con-

taining 5 mg. of phosphatide were hydrolysed with 6N-HCI(1 ml.) at 1000 for 15 min. After dilution with water(1 ml.), the solution was filtered and the precipitate washedwith water (1 ml.). The filtrate and washing were evapor-ated to dryness at 500 in vacuo, and the residue was dis-solved in water (0 5 ml.). Portions containing 20,g. ofphosphorus were applied to Whatman no. 54 paper andchromatographed in ethyl acetate-pyridine-water asdescribed by Hanes & Isherwood (1949).

Preparation of silicic acid. Silicic acid (500 g., 100-mesh,Mallinckrodt A.R. grade from Savory and Moore Ltd.,Lawrence Rd., London, N. 15) was thoroughly mixed with2 1. of water and the suspension allowed to settle for 1 hr.in a 2 1. measuring cylinder. The supernatant liquid con-taining the finer particles was rejected, and the residue wasfiltered off under suction and dried in an air oven over-night at 105-110°. The material was lightly ground andsieved through a 100-mesh nylon gauze and stored in atightly stoppered bottle.

Characterization of phospholipids. The various phospho-lipid fractions, obtained by column chromatography, wereidentified and characterized by chromatography of theunhydrolysed phospholipids on silicic acid-impregnatedpaper, by the methods of Marinetti, Erbland & Kochen(1957) and Lea et al. (1955). Rhodamine 6G followed byultraviolet light was used to detect all lipids, the ninhydrinreaction to detect the amino lipids and phosphomolybdic

186 1961

BUTTER PHOSPHOLIPIDS AND BLOOD COAGULATIONacid reagent for lipids containing choline. Sphingomyelinwas also characterized by its reineckate reaction, its re-sistance to mild alkaline hydrolysis and the presence of achloroform-extractable nitrogenous base after prolongedhydrolysis with Ba(OH)2 solution followed by acidification.Choline was detected by the method of Levine & Chargaff(1951) and by its reineckate reaction (Glick, 1944).

After hydrolysis of the lipids, the bases liberated werecharacterized by paper chromatography by the methods ofDatta, Dent & Harris (1950). Phosphoric esters weredetected by the method of Hanes & Isherwood (1949), andglycerophosphoric acid was determined by the method ofBurmaster (1946). Sugars were detected by paper chro-matography (Jermyn & Isherwood, 1949).

Fractions containing sugar. The sugar was shown to bepresent in chemical combination as follows. Emulsificationof the individual fraction with water in a sonic vibratorfollowed by Seitz filtration gave a filtrate free of sugar, andfractionation between water and chloroform gave anaqueous layer free of sugar. The sugars were, however,readily liberated by acid hydrolysis as indicated above.

Butter. Numerous samples of butter were used in thiswork, all obtained by retail purchase. No obvious differenceswere noted between different brands bought at variousseasons of the year.

Phospholipids. Synthetic L-oc-dimyristoylphosphatidyl-ethanolamine and L-a-dimyristoyl-lecithin were kindlysupplied by Dr Erich Baer (University of Toronto, Canada).A sample of sphingomyelin was kindly supplied by DrElizabeth Smith (Courtauld Institute of Biochemistry,Middlesex Hospital Medical School, London, W. 1) andothers were purchased from L. Light and Co. Ltd., Coln-brook, Bucks, England.

Thromboplastic activity. The blood-clotting activity of thelipids was tested on normal fasting-human plasma by theStypven method and by plasma and whole-blood recal-cification techniques as described earlier (Maclagan &Billimoria, 1956; Billimoria et al. 1957). The thrombin- andthromboplastin-generation tests used were performed asdescribed by Biggs & Macfarlane (1957). Theweight of lipidadded was obtained by multiplying its P content by 25.

Apparatus. A graduated cylindrical donor vessel diam.60-6 mm. containing methanol (1-5 1.) was connected at thebase by means of a stopcock (Sl) to a similar receiver vessel(B) diam. 83-5 mm. containing CHCl3 (1-5 1.). The columncontaining silicic acid was connected by means of a stop-cock (S2) attached to a second outlet in (B) and the fractionswere collected in a fraction collector. Vessel (B) carrieda small iron nail covered in a Fluon sleeve which, in con-junction with a rotating magnet held under the receiver,acted as a stirrer.The diameters of the vessels were such that when, with

the stirrer in motion, the stopcock (S2) was closed and (S,)opened, there was no solvent flow from one cylinder to theother. The gradient elution was started by opening thestopcock (S2).The gradient obtained was checked by introducing a

standard dye solution of Sudan Black in methanol in thedonor vessel and reading the colour of the resulting mixtureof CHCl3-methanol-dye issuing from the receiver vesseland collecting in the fraction cutter tubes.The gradient could also be obtained by a simple calcula-

tion without performing the above experiment. If H1 isthe height of the methanol and H2 the height of the mixture

after a volume V has been collected, then H1/H2 = P1/P2,where Pi and P2 are the densities of methanol and mixturerespectively. The density of the mixture (P2) can be foundafter a fixed volume V leaves the cylinder (R) from measure-ments of H1 and H2. Since the change in volume due to themixing of CHCl3 and methanol was found to be negligible,the percentage composition of the two solvents may beread off directly from a reference line drawn joining thetwo points corresponding to the density of pure CHCl3 andpure methanol.

RESULTS

Dietary fats and Stypven times

The Stypven clotting times of plasma were sub-stantially decreased by adding certain fats toplasma. Thus Fig. 1 shows the effect of two typicalfats, indicating that butter is much more effectivethan margarine at all concentrations tested. Themaximum effect was produced by 10 mg. ofbutter/system, which produced a 78% drop inclotting time.

Clotting unit. Since butter was the most activenatural fat tested, the activity of the least amountof butter (10 mg.) which produced the maximumshortening of clotting time was chosen as oneclotting unit. The other fats could then be assignedunits/mg. With feebly active fats or certain lipidfractions available only in limited quantity, it wasimpossible to produce minimum clotting times andthe units were read from the calibration curvebased on a serial dilution of butter. Thus if0 01 mg. of a substance gave a clotting time of11 sec. with the plasma used in Fig. 1, correspond-ing to 1 mg. of butter (0.1 unit), this substancewould have an activity of 10 units/mg.

40

-70O 30Ca

'544. 200CD

*- 10bo0

0

Conen. of fat (mg./system)

Fig. 1. Effect of adding butter or margarine on theStypven clotting times of normal fasting plasma. *,Butter; 0, margarine; x, control.

Vol. 78 187

J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGAN

Other clotting 8ystem8

The effect of various concentrations of butterand margarine on the plasma recalcification systemwere also studied; Fig. 2 shows that again butterwas far more active than margarine. Maximumeffects were obtained from 5 mg. of butter and a

much smaller maximum from 1 mg. of margarine.The inhibitory effects obtained with larger quanti-ties of the fats were quite pronounced with thisclotting system, especially with margarine, whichgave a clotting time longer than the control at a

concentration of 10 mg./system. This latter effectis not clearly understood but may be due tomechanical interference of neutral fat with theclotting system.

Butter was also tested in the whole-bloodcoagulation system, and the thrombin- and throm-boplastin-generation tests; it was active in allthree systems.

Fractionation of butter

A typical preliminary separation of activematerial from butter is shown in Fig. 3, where theprogress of fractionation was followed by estima-tion of clotting activity at various stages. Butter(1) was melted overnight at 400 and centrifuged(2000 g for 0-5 hr.), when four visible layers were

formed. The top, clear, neutral-fat layer (2) was

almost inactive; this was followed by three aqueouslipoprotein layers which contained the majorproportion of the activity. These aqueous layers (3)were combined and extracted with ether at 400 toremove most of the residual neutral fat. In thisprocess only 10% of the activity was found in the

_ 250C)0

C 200

.5

o

V 150

10-4 10 -3 10 -2 10 -1 1 10

Conen. of fat (mg./system)

Fig. 2. Effect of adding butter or margarine on the recalci-fication times of normal fasting plasma. 0, Butter; 0,

margarine; x, control.

extract (4), even when chloroform was used insteadof ether, suggesting that the active lipid is presentmostly in the form of lipoprotein at this stage. Inconfirmation of this, when the ether-insolubleresidue (5) was treated with boiling acetone andthe mixture filtered after cooling, the activity ofthe dry, acetone-insoluble precipitate (7) wasalmost quantitatively extracted by chloroform(Soxhlet extraction). The chloroform extract (9)was evaporated to a small bulk, and the phospho-lipids (10), precipitated at 00 with a large volumeof acetone, had an activity of approx. 40 units/mg.This represented a 400-fold concentration and anoverall recovery of 74% of the total originalactivity. The phospholipids, dissolved in a mini-mum of chloroform, were adsorbed on a shortsilicic acid column, and the column was washedwith chloroform to remove non-phospholipidfractions. The crude phospholipids (12) wereeluted with ethanol-chloroform-water (6:3:1, byvol.) and, after evaporation of the solvent, had anactivity of approx. 60 units/mg. The acetone-soluble fractions (6) and (11) and the residue in-soluble in chloroform (8) were almost inactive.Having once established that the activity from

butter was due to its phospholipid content, forfurther work, the chloroform extracts from stage(9) were directly adsorbed on a column of silicicacid and concentrated as indicated at stage (12).Thus, in large-scale work, extracts of as much as15 g. of total lipid containing about 1 g. of phos-pholipid could be directly adsorbed on large-diameter columns (2j in.) containing silicic acid(125 g.), and the lipids other than phospholipidseluted off the column with chloroform (usuallyabout 21.). The phospholipids were then eluted offthe columns with the mixture of ethanol-chloro-form-water. After evaporation of this solvent, thelipid, dissolved in a small volume of dry chloro-form, was adsorbed on to a second column forstepwise or gradient elutions, as described below.

StepWi8e elution of butter pho8pholipid8from a 8ilicic acid column

The above phospholipids, dissolved in a smallvolume of dry chloroform, were readsorbed on acolumn of silicic acid (125 g.) and eluted withgraded amounts of methanol, 1, 2, 4, 8, 16 and32 % (v/v) in chloroform and finally with a mixtureof ethanol-chloroform-water (6: 3: 1, by vol.). Theseparation was followed by weighing the lipidseluted and determining their clotting activity. Ateach change of methanol concentration a sharppeak occurred, but the analytical data on theeluates showed that most of the substances ob-tained were still mixtures, and consistent resultswere unobtainable from repeated experiments.

188 1961

BUTTER PHOSPHOLIPIDS AND BLOOD COAGULATION

Butter (441 g., 100 u./g., 44100 t.u.)was melted at 400, centrifuged

and clear fat decanted

2 Neutral fat335 g.

0-06 u./g.20 t.u.

3

4

Lipoprotein layers andresidual neutral fat

extracted twice with ether(0.5 1.) at 40 and centrifuged

l 1Neutral fat removed 5 Lipoproteins

33 g. (ether removed in132 u./g. vacuo)4356 t.u. 73 g.

500 u./g.36500 t.u.

5g. withdrawnf (-2500 t.u.)

68g. was treated with boiling acetone(200 ml.),cooled to 40 and 500 ml. ofacetone added; mixture was cooledto 40 for 1 hr. and filtered

Acetone insoluble142 5 g.

2000 u./g. 2g ihrw29000 t.u. 2gwihrn--"(-4000 t.u.)

9 CHCI3-soluble extract wa15 ml., diluted with 75 macetone, left overnight at

Acetone-CHCI3-soluble1 29 g.2 u./g.2 6 t.u.

Residue(12.5 g.)wasextracted with CHCI3 (400 ml.,8 hr., Soxhlet)

,s evaporated to 8 Residue, insolubleI. of in CHCI340 and filtered 1-2 u./g.

I 12-8 t.u.

10 Acetone-CHCI3-i nsol u ble540 mg.39120 u./g.21130 t.u.

12

J-1-

Fig. 3. Preliminary separation of phospholipids from butter. u, Stypven clotting units;t.u., total units of clotting activity.

1

6

11

Adsorbed on silicic acid (125 g.)column in dry CHCI3, washed withCHCI3 and eluted off column with

ethanol-CHCI3-water (6: 3: 1, by vol.)to give the phospholipids

360 mg.60000 u./g.21 600 t.u.

Vol. 78 189

J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGANElementary chemical analysis of the lipids and

paper chromatography of intact lipids and theirhydrolysis products show that the 8 % fractioncontained mainly serine, glycerophosphoric acidand galactose; ethanolamine was also present to alesser extent in the hydrolysate. The fraction had aclotting activity of approx. 200 units/mg. Theanalysis (Found: C, 616;H, 10;N, 1.5;P,37%)was in close agreement with that of a phospha-tidylserine galactoside containing C18 and C20 acids,with an empirical composition C49H.5014NP (re-quires C, 61*8; H, 100; N, 1 5; P, 3-4%).The 32 % fraction was of the lecithin type since

it gave no ninhydrin reaction even after hydrolysisand contained choline, glycerophosphate, glucoseand galactose. The P:N ratio was 1: 1- 14 and thelow carbon value was due to the glycoside struc-ture (Found: C, 57-1; H, 101; N, 19; P, 3-7%).The fraction was inactive in the clotting test.The ethanol-chloroform-water fraction con-

tained mainly sphingomyelin as shown by paperchromatography, precipitation reaction with Rei-necke's salt, and presence of choline and absence ofglycerophosphate in the hydrolysate. The P :Nratio was 1:2'14 (Found: C, 57 9; H, 10-3; N, 2-9;P, 3 0 %). The 1, 2, 4 and 16% fractions appearedto be mixtures and were not further investigated.

Gradient elution chromatography of phospholipid1Further information was obtained by elution of

the lipids from a silicic acid column by a lineargradient elution technique; the crude phospho-lipids, after removal of neutral fat, were dissolvedin dry chloroform and adsorbed on a column ofsilicic acid. Fig. 4 shows the separation of thelipids on elution with a continuous linearly in-creasing concentration of methanol in chloroformso adjusted that after 108 fractions of 25 ml. hadbeen collected, the solvent in the mixing chamberwas approx. 90% of methanol and 10% of chloro-form. Each fraction of the eluate was analysed forphosphorus, total nitrogen and sugar; these areplotted as fLg.atoms of P and N, and ,umoles ofsugar per fraction. The clotting activities were alsodetermined.The earlier fractions 17-49 which constitute the

the first five peaks were poorly separated and afteracid hydrolysis were found to contain over 90%of the total nitrogen as amino nitrogen; thesefractions were further investigated after rechro-matography. Fractions 50-108 contain choline asthe base and were free from amino nitrogen. Theresults of rechromatography of these fractions isshown below.

70

60 °1-

v

0

50 A0

0.0

4)

40

03030.0

CD

C.0

20 ~i0

00

10

20 30 40 50 60 70 80 90 100Fraction no. (25 ml. fractions)

Fig. 4. Chromatography of butter phospholipids by a continuous linear gradient elution with methanol-CHC13from a column of silicic acid. 0, Phosphorus; 0, nitrogen; x, sugar; -, slope of the solvent gradient. Peaknumbers are in parentheses.

I 3-50

. 4

C)

s, 3.0P4m4)05 2-5

0

20bo

Om 2-0

'd0

"0

P0 Q.5

0

190 1961

BUTTER PHOSPHOLIPIDS AND BLOOD COAGULATION

Peak 6 (fractions 50-59, Fig. 4), lecithin galacto-side. The intact lipid was detected on a silicic acidpaper chromatogram with Rhodamine 6G and thephosphomolybdic acid reagent; its R. was 0 44.The P: N: sugar ratios were approx. 1: 1: 1, andafter hydrolysis the products identified wereglycerophosphoric acid, choline and galactose. Thelipid had no clotting activity.Peak 7 (fractions 70-79, Fig. 4), a sugar-containing

phosphatidylcholine. The lipid was detected bysilicic acid paper chromatography with Rhodamine6G and the phosphomolybdic acid reagent. TheP:N ratio after reprecipitation (1 vol. of chloroformplus 10 vol. of acetone) was 1:1-2. After hydro-lysis, galactose and glucose were detected, and thelipid had a clotting activity of 50 units/mg.Peak 8 (fractions 82-89, Fig. 4), lecithin. The

intact lipid on silicic acid paper chromatographywas detected with Rhodamine 6G and phospho-molybdic acid; its RF, 0-71, was identical with thatof synthetic lecithin run alongside and in ad-mixture. The P:N ratio, after reprecipitation asabove, was 1:108 (theory requires 1:1) and acid-hydrolysis products were choline and glycerophos-phate. The lipid had no clotting activity.Peak 9 (fractions 100-108, Fig. 4), sphingomyelin.

The intact lipid on silicic acid paper chromato-graphy was detected with the Rhodamine 6Gand phosphomolybdic acid reagents, R. 0-61. Itreadily gave a reineckate and had a P:N ratio of1:2-18 (theory 1:2). On acid hydrolysis followedby chromatography (as described for phosphateesters) the phosphate ester had R, 0 8 (withreference to glycerophosphate RF 1). Choline wasliberated on prolonged acid hydrolysis, as de-scribed under hydrolysis for bases. The lipid had noclotting activity.

Rechromatography of peaks 1-5

Fractions 17-50 (Fig. 4), after removal ofsolvent, were dissolved in a small volume of chloro-form, rechromatographed on a similar silicic acidcolumn and eluted with a linear gradient ofmethanol in chloroform, so adjusted that after60 fractions of 25 ml. a concentration of 30% (v/v)of methanol in chloroform was reached in themixing vessel. The separation of the early peaks(Fig. 5) was considerably improved, especially withrespect to the sugar peak, and the nitrogen andsugar values of the fractions ran closely parallelwith the phosphorus values. The peak fractionafter reprecipitation as above gave the followingresults.Peak 1 (Fig. 5), a complex phosphatidic acid. On

silicic acid paper chromatography this gave noninhydrin reaction and was detected with Rhod-amine 6G and phosphomolybdate reagents, R. 0 95(considerable 'tailing'). The N content was neg-ligible, and on hydrolysis no base was detectableand glycerophosphoric acid was obtained (tailingon chromatogram). The lipid had a weak clottingactivity of 5 units/mg.Peaks 2, 3 and 4 (Fig. 5), phosphatidylserine and

phosphaticlylserine galactoside. On silicic acid paperchromatography these peaks gave ninhydrin-positive spots which were also detectable withRhodamine 6G (violet fluorescence under u.v.light) at RF 0 73, 0-72, 0-72 respectively. The P:Nratio of each of the compounds was 1 :0-87, 1: 11and 1: 0 88 respectively, whereas peak 4 containedgalactose (P: galactose ratio, 1: 4 5). Hydrolysis ofpeaks 2 and 3 gave glycerophosphoric acid andserine whereas hydrolysis of peak 4 gave glycero-phosphoric acid, serine and galactose. The clotting

- 20E0

-4

; 1o5

0

zo~

CCa

0

300

200

10 I-

0

S

000

20 30 40 50 60Fraction no. (25 ml. fractions)

Fig. 5. Rechromatography of peaks 1-5 from Fig. 4 with gradient of methanol increasing to 34% after70 fractions. *, Phosphorus; 0, nitrogen; x , sugar; - , slope of gradient. Receiver diam. 83-5 mm. and donor73 mm. Concentration of methanol in CHC13 in donor vessel, 57% (v/v).

Vol. 78 191

J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGAN

activities of peaks, 2, 3 and 4 were 8t30 units/mg. respectively.

Phosphatidylserine appears as twofractions (peaks 2 and 3) when run withsolvent system whereas both compounctogether when chromatographed on sipaper with acid solvents.Peak 5 (Fig. 5), phosphatidylethanolam

phatidylserine trace8, compound X. The lijfraction appears to be the least homogeron silicic acid paper chromatography gdistinct components. (1) Phosphatid3amine, R. 0-76, which gave a positive:reaction and with Rhodamine 6G a yelescent spot when viewed under u.v. liglentic L - a - dimyristoylphosphatidylethEgave identical reactions and had R.(2) Phosphatidylserine, R. 0-72, gave a npositive reaction and blue fluorescenceRhodamine 6G reagent when viewed ulight. (3) Compound X, with R. 0.54negative reaction with ninhydrin andmolybdic acid reagent, but it appearedfluorescent spot with the Rhodamine 61under u.v. light. The mixed lipids showedactivity of 500 units/mg. The mixture (

lysis showed the presence of ethanolaserine.The mixture of phosphatidylethanola

phosphatidylserine was separated from (

X by cutting the paper. After elutionclotting activity/mg. of P was 100 timeas that of the mixed phosphatidylethanolphosphatidylserine fraction.

Table 1. Thromboplastic activitof butter phospholipids

Peaks 1-5 refer to Fig. 5 and 6-9 to Fig. 4. C[5 (c)] was approx. 100 times as active (per mg.mixture of 5 (a) and 5 (b).

Peak no.

1234

PhospholipidA complex phosphatidic acidPhosphatidylserinePhosphatidylserine

Phosphatidylserine galactoside5 Mixture:

(a) Phosphatidylethanolamine(b) Phosphatidylserine(c) Unknown compound X

6 Phosphatidylcholine galactoside7 A choline lipid containing glucose

and galactose8 Lecithin9 Sphingomyelin

Synthetic L-a-dimyristoylphospha-tidylethanolamine

5, 60 and

I distincta neutralIs appearLicic acid

mine, phos-pid in thisaeous and,ave threeylethanol-ninhydrinlow fluor-

Summary of phospholipid fractions andtheir blood-clotting activities

A list of the phospholipids separated from butteris shown in Table 1, together with their clottingactivities. It will be seen that the strongestclotting activity was in fraction 5 and, as shownabove, the most active in this group is compound X.The synthetic L-c-dimyristoylphosphatidylethanol-amine which was tested at the same time on thesame plasma specimen exhibited only a weakactivity.

DISCUSSION

ht. Auth- Although much has been published on thenolamine analysis of butter and dairy fats (reviewed by0-74-76. Hilditch, 1949; Davis & Macdonald, 1953), little

iinhydrin- attention has been directed to the butter phospho-with the lipids which are described here. We have foundmder u.v. that the whole of the strong thromboplastic4, gave a activity of butter resides in a number of the phos-phospho- pholipid factions and these appear to be present asas a blue lipoproteins in the original butter. Ten phospho-G reagent lipids have been separated and six of these havea clotting thromboplastic activity. The most active, com-on hydro- ponent X, which has been recognized only by,mine and silicic acid paper chromatography, remains un-

identified; it was approximately 100 times astmine and active as phosphatidylethanolamine and phospha-compound tidylserine prepared from butter and far moreof X, its active than synthetic phosphatidylethanolamine.

s as great During the progress of this work Poole & Robin-amine and son (1956) and O'Brien (1956a) reported blood-

thromboplastic activity from phospholipid frac-tions of egg and, when Lea et al. (1955) isolatedphosphatidylethanolamine from the same source,

Y the activity of egg yolk was attributed to itsphosphatidylethanolamine content. In our ex-

!ompound X perience egg yolk was only weakly active comparedof P) as the with butter, and as butter contains only traces of

Clotting phosphatidylethanolamine its activity is unlikelyaCtivity to be due to this substance. It appears from the

(units/mlg.) evidence presented here that phospholipids other5 than phosphatidylethanolamine are strong potenti-85

ators ofblood coagulation. The difficulty ofassessing85 the activity of thromboplastic agents has been60 overcome by the use of an arbitrary system of30 blood-clotting units where butter has been chosen

as a standard of comparison, not only because it

500 was the most active fat, but also because a largenumber of samples of butter tested had approxi-

0 mately the same thromboplastic activity.50 Silicic acid chromatography has long been

recognized as a powerful tool in the separation of0 phospholipids and although various techniques0 have been reported with this absorbent, it appears7 that the choice of the method of elution is de-

pendent on the composition of the lipids requiring

192 1961

Vol. 78 BUTTER PHOSPHOLIPIDS AND BLOOD COAGULATION 193

separation (Lea et al. 1955); Hanahan, Dittmer &Warashina, 1957; Marinetti, Scaramuzzino &Stotz, 1957). With the Lea et al. method the butterphospholipids were eluted as a single peak, andfurther experience indicated the need for a gradientelution system.When a modified exponential system of elution,

as used by Alm, Willias & Tiselius (1952) for theseparation offatty acid mixtures, was used with themethanol-chloroform solvent system, the concen-tration of the polar solvent increased too rapidly inthe early stages of chromatography. The lineargradient chromatography eventually used in thiswork is a modification of the systems of Kellie &Wade (1957) and Edwards (1958). The largedifferences in the densities of methanol and chloro-form necessitated the use of donor and receivervessels of different cross-sectional areas. With sucha gradient elution system, highly reproduciblechromatographic separation was obtained andpreliminary experiments on the separation ofphospholipids from egg yolk and blood plasmasuggest that the method may be of general use inthe fractionation of phospholipids.The fatty acid composition of the phosphoipide

has not been investigated and the nature of thesugar linkage in the glycosides is uncertain,although it has been established that the sugar isnot present in a free state and is probably linkedthrough the amino nitrogen in the 'kephalin'group of lipids. Phosphatidylserine appears astwo distinct peaks (peaks 2 and 3, Fig. 5) and it isunlikely that separation within this group is ob-tained through a difference in fatty acid composi-tions, for on silicic acid paper chromatography inacid media the two components run together. Thisobservation is in agreement with that of Marinetti,Erbland & Stotz (1958), who found two ionic formsof phosphatidylserine in brain phospholipids.The physiological importance of the butter

phospholipids described is uncertain since there isevidence that such compounds are probably modi-fied during intestinal absorption (Ahrens &Borstrom, 1956; O'Brien, 1957; Maclagan, Billi-moria & Curtis, 1958). Nevertheless, the possibleinfluence of dietary phosphoipids on blood phos-pholipids is of some interest and is under furtherinvestigation.

SUMMARY

1. The addition of butter to plasma shortenedthe Stypven clotting time ofhuman plasma by over75% and butter was more active than any otherfat tested.

2. A preliminary separation of butter lipidsindicated that the thromboplastic activity wasconfined to the phospholipids originally bound tothe proteins of butter.

3. Butter phospholipids were separated bychromatography on silicic acid columns by step-wise and continuous gradient elution techniques,with methanol-chloroform mixtures. A simplemethod of obtaining an approximately lineargradient is described which may be of general usein the separation of phospholipid mixtures.

4. Ten major phospholipids have been detectedin butter and most of these have been identified bypaper chromatography and chemical analysis ofthe intact lipids and their hydrolysis products.

5. A comparison of the Stypven clotting activi-ties of the butter-phospholipid fractions with thatof synthetic L-ac-dimyristoylphosphatidylethanol-amine has shown that a number of these fractionsare far more active than the synthetic compound.The most active butter phospholipid (compoundX), has not yet been identified.

We are indebted to Miss Sheila Pateman for valuabletechnical assistance. The work was supported by generousgrants from the endowment funds ofWestminster Hospital.

REFERENCES

Ahrens, E. H. & Borstrom, B. (1956). J. biol. Chem. 219,665.

Allen, R. J. L. (1940). Biochem. J. 34, 858.Alm, R. S., Williams, R. J. & Tiselius, A. (1952). Acta

chem. 8cand. 6, 826.Biggs, R. & Bidwell, E. (1957). Brit. J. Haematol. 3, 387.Biggs, R. & Macfarlane, R. G. (1957). Human Blood

Coagulation and it8 Disorders, 2nd ed. Oxford: BlackwellScientific Publications Ltd.

Billimoria, J. D. & Curtis, R. G. (1959). Biochem. J. 72,29P.

Billimoria, J. D., Lockey, E. & Maclagan, N. F. (1957).Proc. Roy. Soc. Med. 50, 623.

Burmaster, C. F. (1946). J. biol. Chem. 164, 233.Datta, S. P., Dent, C. E. & Harris, H. (1950). Science, 112,

621.Davis, J. G. & Macdonald, F. J. (1953). In Richmond's

Dairy Chemistry, 5th ed. London: Charles Griffin andCo. Ltd.

Edwards, R. W. H. (1958). In Chromatographic Techniques,Ed. by Smith, I. London: William Heinemann MedicalBooks Ltd.

Fullerton, H. W., Davie, W. T. A. & Anastasopoulos, G.(1953). Brit. med. J. 2, 250.

Glick, D. (1944). J. biol. (Chem. 156, 643.Hanahan, D. J., Dittmer, J. C. & Warashina, E. (1957).

J. biol. Chem. 228, 685.Hanes, C. F. & Isherwood, F. A. (1949). Nature, Lond.,

164, 1107.Hilditch, T. P. (1949). Chemical Constitution of Natural

Fat8, 2nd ed. London: Chapman and Hall Ltd.Jermyn, M. A. & Isherwood, F. A. (1949). Biochem. J. 44,

402.Kellie, A. E. & Wade, A. P. (1957). Bioch,em. J 66,196.King, E. J. & Wootton, I. D. P. (1956). Micro-Analysis in

Medical Biochemistry, 3rd ed. London: J. and A.Churchill Ltd.

13 Bioch. 1961, 78

194 J. D. BILLIMORIA, R. G. CURTIS AND N. F. MACLAGAN 1961

Lea, C. H. (1956). Proc. 2nd int. Conf. Biochem. Probl.Lipids (Ghent, 1955), p. 81.

Lea, C. H. & Rhodes, D. N. (1954). Biochem. J. 56, 613.Lea, C. H., Rhodes, D. N. & Stoll, R. D. (1955). Biochem. J.

60, 353.Levine, C. & Chargaff, E. (1951). J. biol. Chem. 192, 465.Macfarlane, R. G. (1955). Brit. med. BuU. 2, 1.Macfarlane, R. G., Trevan, J. W. & Attwood, A. M. P.

(1941). J. Phy8iol. 99, 7i.Maclagan, N. F. & Billimoria, J. D. (1956). Lancet, ii, 235.Maclagan, N. F., Billimoria, J. D. & Curtis, C. (1958).

Lancet, ii, 865.Marinetti, G. V., Erbland, J. & Kochen, J. (1957). Fed.

Proc. 16, 837.

Marinetti, G. V., Erbland, J. & Stotz, E. (1958). Biochim.biophy8. Acta, 30, 41.

Marinetti, G. V., Scaramuzzino, D. J. & Stotz, E. J. (1957).J. biol. Chem. 224, 819.

O'Brien, J. R. (1956a). J. dlin. Path. 9, 47.O'Brien, J. R. (1956b). Lancet, ii, 232.O'Brien, J. R. (1957). Lancet, i, 1213.Poole, J. C. F. & Robinson, D. S. (1956). Quart. J. exp.

Phy8iol. 41, 295.Poole, J. C. F., Robinson, D. S. & Macfarlane, R. G.

(1955). Quart. J. exp. Phy8iol. 40, 276.Robinson, D. S. & Poole, J. C. F. (1956). Quart. J. exp.

Physiol. 41, 31.Winzler, R. J. (1955). Meth. biochem. Anal. 2, 290.

Biochem. J. (1961) 78, 194

On the Mechanism of et-Oxoglutarate Oxidation in Escherichia coli

BY L. P. HAGER* AND H. L. KORNBERGMedical Research Council Cell Metabolism Research Unit, Department of Biochemistry,

University of Oxford

(Received 28 June 1960)

The enzymic formation of succinyl-coenzyme Afrom a-oxoglutarate is known to require diphos-phopyridine nucleotide (DPN+), thiamine pyro-phosphate (TPP), lipoic acid (lip-So.. or lip-S,,d.)and coenzyme A (CoA). It has been postulated(Gunsalus, 1953; Kaufman, 1955; Sanadi, Langley&; White, 1959) that the sequence of reactions in-volved in the overall reaction is:

(1) a-oxoglutarate +TPP - [succinic semialdeyde]-TPP + C02;(2) [succinic semialdehyde]-TPP + lip-S... = succinyl-S-lip + TPP;(3) succinyl-S-lip + CoA-SH = succinyl-S-CoA + lip-SrYd.;(4) lip-Sr,d +DPN+ = lip-So. + reduced DPN+ H+.

Sum:

(5) oa-oxoglutaxate+DPN+ + CoA-SH -I_succinyl-S-CoA+ reduced DPN+H+ + C02 .

Attempts to isolate the enzymes catalysing theseindividual steps have been largely unsuccessful,although lipoic dehydrogenase (reaction 4) has beenseparated and shown to be a component of the a-oxoglutarate-dehydrogenase system of E8cherichiacoli (Hager & Gunsalus, 1953). More recently themammalian oc-oxoglutarate-dehydrogenase systemhas also been resolved with respect to lipoic de-hydrogenase (Massey, 1960) and it has been shownto be a flavoprotein (Massey, 1958; Searls & Sanadi,

1959) identical with Straub's (1939) diaphorase(Massey, 1960).

It is the purpose of this paper to characterize thebiochemical lesion in a succinate-requiring mutantof E. coli. The experiments reported indicate thatthe mutant lacks the enzyme catalysing reaction 1,cx-oxoglutarate carboxylase, or reactions 1 and 2.It therefore provides a source for the a-oxoglutar-

ate-dehydrogenase complex which is deficient inthis enzyme or these enzymes.

METHODS AND MATERIALS

Maintenance of 8tock culture. E. coli W (wild type) andthe succinate-requiring mutant (E. coli 309-1 R) were giftsfrom Professor B. D. Davis (Harvard University MedicalSchool, Boston, Mass., U.S.A.). The organisms were main-tained in stock culture on agar slopes containing (per100 ml. of medium) 1 g. of tryptone, 1 g. of Difco yeastextract, 0 5 g. of dipotassium hydrogen phosphate, 0'3 g.of glucose, 0-2 g. of sodium suceinate and 2 g. of agaragar (Hopkin and Williams Ltd., Chadwell Heath,Essex).

* Fellow of the John Simon Guggenheim MemorialFoundation. Present address: Department of Chemistryand Chemical Engineering, University of Illinois, Urbana,Ill. (U.S.A.).